报告详情
  1. 活动首页
  2. 报告清单
  3. 报告详情
A Perception-Reasoning-Planning Framework for Vision-Guided Robotic Assembly of Industrial Components
编号：26 访问权限：仅限参会人 更新：2026-09-14 12:23:14 浏览：1次 口头报告

报告开始：暂无开始时间（Asia/Shanghai）

报告时间：暂无持续时间

所在会场：[暂无会议] [暂无会议段]

暂无文件

摘要
High-success-rate robotic assembly of industrial components requires stable perception, accurate pose estimation, and smooth motion execution under complex operating conditions. However, conventional vision-guided systems are often affected by illumination variations and depth noise, resulting in unstable localization. Furthermore, traditional motion planning often overlooks kinematic constraints, causing abrupt dynamic disturbances. This paper presents an integrated perception-reasoning-planning framework to improve the performance of vision-guided robotic assembly and increase its success rate. In the perception stage, an Interacting Multiple Model Kalman Filter (IMM-KF) adaptively fuses visual observations across different motion states to improve the temporal consistency of three-dimensional localization. For axisymmetric components, a geometry-aware reasoning method is established by combining point cloud preprocessing, Random Sample Consensus (RANSAC)-based end-face extraction, and Principal Component Analysis (PCA)-based normal estimation, enabling geometry-based 6-DoF pose estimation without deep regression training. Finally, a hierarchical motion planning strategy is incorporated, utilizing manipulability-based singularity avoidance and second-order continuous (C²) trajectory generation to improve motion execution stability. Experimental results on a collaborative robotic platform performing shaft-hole assembly under varying component placement angles and positions validate that the proposed framework enhances spatial localization consistency and significantly increases the assembly success rate from 32% to 89% under practical assembly uncertainties.
 
关键词
Vision-guided robotic assembly, Geometric reasoning, 6-DoF pose estimation, Motion planning, Industrial robot
报告人
Mengyang Zhang
Master Student Xi'an Jiaotong University

稿件作者
Mengyang Zhang Xi'an Jiaotong University
Zeqi Wei Xi'an Jiaotong University
Ruqiang Yan Xi'an Jiaotong University
发表评论
验证码 看不清楚，更换一张
全部评论
重要日期

  • 会议日期

    11月06日

    2026

    至

    11月08日

    2026

  • 10月15日 2026

    初稿截稿日期

主办单位
IEEE Instrumentation and Measurement Society
承办单位
Sichuan University
联系方式
登录查看完整联系方式
历届会议
移动端
在手机上打开
小程序
打开微信小程序
客服
扫码或点此咨询